Detecting change in real-world scenes

Introduction People tend to be surprisingly poor at detecting changes in visual images of real-life situations. Such change blindness suggests a limitation to the human capacity to encode and compare visual information from one moment to the next, especially in images fol-lowing a large transient or flicker. We therefore used the natural indoor scenes of the change blindness database by Sareen and colleagues (2015), containing the size and on-screen location information of all changes in the images, within a flicker task paradigm. Method In this task an original and a modified real-world scene, separated by a grey blank, al-ternated repeatedly until observers detected the location (left or right side) of the im-plemented change. The stimuli were either presented upright or inverted. The goal of the study was twofold: (1) using change size and location as predictor of participant performance and (2) identifying the influence of image inversion in change detection. Results A random intercepts (logistic) regressions analysis for RT and accuracy revealed three significant main effects: (1) image inversion (upside-down worse and slower than nor-mal image presentation), (2) object size (larger objects easier and faster than smaller ones) and (3) location (peripheral changes easier and faster than central ones). Discussion Change size and location had a profound impact on search performance. This might relate to the overall saliency of the induced change and the search strategy of the partic-ipants. Performance was also influenced by scene inversion, possibly due to a less meaning-driven endogenous search strategy when scenes were presented upside-down.status: publishe